The present research sought to explore the implementation of the PEAK Relational Training System’s Generalization and Direct Training modules in a practical setting across two studies. Study 1 investigated whether a relationship exists between the PEAK-G and PEAK-D Training protocols, if the skills mastered through these would generalize to the natural environment, and whether staff would rate these protocols as socially valid. In Study 1, participants included 12 children aged 4-5 years, 5 with diagnosed autism, 7 with neuro-typical development, and 6 members of staff. Assessments were conducted with all participants using PEAK-G and PEAK-D Modules; subsequently 5 target relational skills from the PEAK-G were taught, followed by a second PEAK-G and PEAK-D assessment. Results showed that following PEAK-G training, children derived target skills in the PEAK-D module that were absent in the first assessment which involved no exposure to PEAK-D training. Mastered PEAK skills were also found to have generalised to the appropriate natural environment. Staff rated these PEAK modules as socially valid and acceptable language training protocols. Study 2 aimed to partially replicate results from Study 1 using a multiple-probe design across behaviours to explore if PEAK-G training would result in related PEAK-D scores being mastered without PEAK-D training. Participants were 4 children aged 4 with diagnosed autism. Baseline conditions involved probes of targeted skills from the PEAK-G and PEAK-D modules. This was followed by PEAK-G training and subsequent tests of related target skills in PEAK-D to determine if learning PEAK-G relational skills resulted in acquisition of PEAK-D relational skills in similar target areas. Results showed that relational PEAK-D target skills that were absent prior to PEAK-G training were evident for 5 children with autism. Procedural fidelity measures also indicated that high levels of fidelity were maintained by staff over an extended period following training via Behaviour Skills Training. Implications of results and suggestions for future research in refining the efficiency of PEAK training are discussed.